Last active
November 9, 2022 10:28
-
-
Save quan-nguyen2/2c0964859b97197a8bc741a23a0e3f36 to your computer and use it in GitHub Desktop.
this script dumps kafka topics to files by day
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| #!/usr/bin/env bash | |
| set -e | |
| if ! command -v kafka-run-class &>/dev/null; then | |
| if ! command -v kafka-run-class.sh &>/dev/null; then | |
| echo "kafka-run-class or kafka-run-class.sh could not be found in PATH" | |
| exit 1 | |
| else | |
| sh=1 # this is most likely we downloaded kafka binary and use *.sh files to working with kafka | |
| fi | |
| else | |
| sh=0 # this is when we install kafka by brew | |
| fi | |
| # extract offsets (start, end) offsets for each partitions of a given topic; print to output and write to output_file | |
| # example: extract_offsets $brokers $topic 2022-11-07 | |
| function extract_offsets() { | |
| local -r brokers=$1 | |
| local -r topic=$2 | |
| local -r day=$3 # something like 2022-07-01 | |
| local -r end_date=$(date -I -d "$day + 1 day") | |
| local -r ts1=$(date -d "$day" +"%s000") # convert 2022-07-01 => timestamp in millisecond | |
| local -r ts2=$(date -d "$end_date" +"%s000") # convert 2022-07-02 => timestamp in millisecond | |
| local -r tmp_dir=$(mktemp -d) | |
| # for debug | |
| if [ $sh -eq 1 ]; then | |
| kafka-run-class.sh kafka.tools.GetOffsetShell --broker-list $brokers --topic $topic --time $ts1 >$tmp_dir/start | |
| kafka-run-class.sh kafka.tools.GetOffsetShell --broker-list $brokers --topic $topic --time $ts2 >$tmp_dir/end | |
| else | |
| kafka-run-class kafka.tools.GetOffsetShell --broker-list $brokers --topic $topic --time $ts1 >$tmp_dir/start | |
| kafka-run-class kafka.tools.GetOffsetShell --broker-list $brokers --topic $topic --time $ts2 >$tmp_dir/end | |
| fi | |
| # join above 2 files | |
| # awk -F':' '{print $(NF-1),"\t",$NF}' write last 2 columns of each file, each column separated by ":" | |
| join <(cat $tmp_dir/start | awk -F':' '{print $(NF-1),"\t",$NF}') <(cat $tmp_dir/end | awk -F':' '{print $(NF-1),"\t",$NF}') | | |
| # print: partition, start_offset, end_offset, end_offset-start_offset | |
| # the result will be something like | |
| # 0 23014 23038 24 | |
| #1 20864 20884 20 | |
| #2 20351 20377 26 | |
| #3 18690 18690 0 | |
| #4 19124 19126 2 | |
| #5 28902 28916 14 | |
| awk '{print $1,$2,$3,$3-$2}' | |
| } | |
| export -f extract_offsets | |
| function dump() { | |
| local -r brokers=$1 | |
| local -r topic=$2 | |
| local -r partition=$3 | |
| local -r start_offset=$4 | |
| local -r end_offset=$5 | |
| if [ $sh -eq 1 ]; then | |
| kafka-console-consumer.sh --bootstrap-server $brokers \ | |
| --topic $topic --partition $partition --offset $start_offset \ | |
| --max-messages $((end_offset - start_offset)) | |
| else | |
| kafka-console-consumer --bootstrap-server $brokers \ | |
| --topic $topic --partition $partition --offset $start_offset \ | |
| --max-messages $((end_offset - start_offset)) | |
| fi | |
| } | |
| export -f dump | |
| # run need brokers, topic, date, output files as result | |
| function run() { | |
| local -r brokers=$1 | |
| local -r topic=$2 | |
| local -r date=$3 | |
| local -r output=$4 | |
| local -r offsets_file=$(mktemp) | |
| local -r temp_out=$(mktemp) | |
| extract_offsets $brokers $topic $date | tee $offsets_file | |
| while read -r line; do | |
| # each line contains 4 fields: partition start_offset end_offset | |
| partition=$(awk '{print $1}' <<<"$line") | |
| start_offset=$(awk '{print $2}' <<<"$line") | |
| end_offset=$(awk '{print $3}' <<<"$line") | |
| echo "dumping partition $partition for offset range [$start_offset, $end_offset]" | |
| dump $brokers $topic $partition $start_offset $end_offset | gzip >>$temp_out | |
| done <"$offsets_file" | |
| mv $temp_out $output | |
| } | |
| export -f run |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment